Exit Rate Calculator

The calculation itself

Exit rate for a page is exits from that page divided by total views of it: exits ÷ pageviews × 100.

It differs from bounce rate, which counts only sessions where the page was the sole page viewed. Every session ends somewhere, so a high exit rate is not inherently a problem.

The useful version compares a page's exit rate against what it should be given its role in the journey.

Running the numbers

An order confirmation page with a 92% exit rate is behaving correctly, the journey is complete.

A shipping information page in the checkout with a 34% exit rate is a problem, since nobody reaching it intended to leave.

On the example store, 1,333 checkouts with a 34% exit at that step is 453 people leaving at the point shipping cost appears. At a 60% probability of conversion had they continued, that is 272 orders and $8,677 of contribution monthly.

The same 34% on a blog post would be unremarkable, which is why the number only means anything against the page's purpose.

What gets missed

Exit rate is frequently confused with bounce rate and the two answer different questions. A page can have a low bounce rate and a high exit rate if people arrive elsewhere, browse to it, and leave.

It is also inflated for any page that legitimately ends a journey, and treating a high figure as a fault on those pages wastes effort.

What to do next

Classify pages by whether exiting is expected before looking at the rate. Confirmation pages, contact pages and content answering a specific question should have high exit rates.

Then investigate only the pages where exiting represents failure: checkout steps, product pages, category pages, and look at what changed rather than at the absolute level.

Session recordings as the next step

Exit rate identifies where people leave and says nothing about why. Session recordings and heatmaps on the specific pages with unexpected exits usually answer the question in an afternoon.

What they typically reveal is mundane: a form field rejecting valid input, a button below the fold on a common screen size, an error message that does not explain what to fix, a third-party script blocking interaction.

Those problems are invisible in aggregate analytics and obvious in a recording, which is why the combination of quantitative identification and qualitative diagnosis is far more productive than either alone. Most sites have two or three of them at any time, and each is worth more than a redesign.

Scroll depth alongside exit rate distinguishes a visitor who left immediately from one who read the whole page and then left, which are entirely different signals about the page.

Most analytics platforms support scroll tracking with a single configuration change and it makes exit data interpretable rather than merely descriptive.

Reviewing exit pages by traffic source separates a page problem from an expectation problem, since visitors arriving from different places want different things from the same page.

Exit rate is only a problem where the page was not meant to be the last one. A high exit rate on an order confirmation page is correct. On a product page it is a signal, and on a checkout step it is urgent. Reading exit rate without asking what the page was for produces a list of problems that mostly are not.